Uploaded by Pieter De Pauw

burndown-book-final2

advertisement
Burndown
A Better Way to Build Products
by:
David Cancel
co-founder & CEO of Drift
Matt Bilotti
Drift product manager
@ 2017 Drift.com, Inc.
All rights reserved.
Designs by: Erik Devaney, Elyse Bogacz, and
Matt Bilotti
Edited by: Erik Devaney
No parts of this book may be used or reproduced in any
manner without written permission from the copyright
owner and publisher, except in the context of reviews.
Foreword
A case study
Let me tell you a quick story about B12, a company that has incredibly high aspirations. They
were working to spin out a new product, but
their current approach to product development just wasn’t cutting it. It was too slow,
and too rigid. They felt trapped in the existing
frameworks that were out there and were experimenting with some new approaches, all of
which couldn’t seem to fill the gap.
Then B12’s CEO, Nitesh, found out about the
Burndown Framework from a series of Medium
posts that we wrote on the topic. They started
by following the advice in those posts and were
eager to learn more, so we set up a time to walk
4 • Burndown
through every different aspect of Burndown.
They kept asking great questions and were
yearning for more detail.
Checking in with them after they got Burndown
up and running within their team, they were
ecstatic to share that The Burndown Framework completely changed their way of dealing
with product and helped them reach (and exceed) their company goals.
In the book below, we explore the philosophy
of Burndown, as well as all the tactical things
you need to do to make it work for you. We’ll
cover everything from the design process, to
handling tech debt, to team structure, and so
much more.
Thanks for picking up this book and we’re excited to dive in with you. Keep reading to learn
how B12 achieved the product development
swiftness they desired.
David Cancel
5 • Burndown
Matt Bilotti
Chapter 1 . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . 7
The Current State of Product Development
Chapter 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
The Burndown Framework
Chapter 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Table of Contents
Getting Started with Burndown
Chapter 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
How to Organize Product Teams
Chapter 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
The Design Process
Chapter 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Customer Research & Creating Artifacts
Chapter 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Dealing with Bugs
Chapter 8 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Tech Debt
Chapter 9 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Managing Your Product Backlog
Chapter 10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Prioritization
Chapter 11 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Transparency & Communication
Chapter 12 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Getting Your Team Onboard
6 • Burndown
Chapter 1
The Current State of
Product Development
No matter how talented your PMs and engineers are, it seems every software company is
still plagued by the following...
• Shipping things too quickly and cutting corners, causing longer-term issues (quick ship
releases)
• Not shipping fast enough and frustrating
sales, success, marketing, support, etc. (big
bang releases)
• Not enough transparency into what the product team is building. Stakeholders feel out of
the loop or unclear on exactly what’s coming or
being worked on
7 • Burndown
• A never-ending game of prioritization and
shifting priorities.
If you’re like us, then no matter what you’ve
tried, these challenges still linger over your
team. But if you’re being honest with yourself,
you know your team can achieve more.
That’s how we felt back in 2009 at Performable, when we first discovered the principles
of Burndown. Then when Performable was acquired in 2011, we got the opportunity to battle-test those principles at HubSpot. Now, at
Drift, we’re perfecting the principles of Burndown and codifying them.
Since implementing the Burndown framework
at Drift, we’ve seen a 5x increase in revenue
- far more growth than we ever thought our
team could achieve. We hope you will see a
similar lift when you finish reading this book.
The Quick Ship & Big Bang Spectrum
Let’s talk about the first two major problems
listed earlier-- shipping too fast or too slow. In
the world of product management, we oscillate
between two ends of a spectrum.
8 • Burndown
Each has its advantages and disadvantages.
On one end, we have “Quick ship” releases,
which are small, highly incremental just-intime releases. On the other end, we have “Big
Bang” releases. These are giant, long undertakings that attempt to have every detail thought
through and perfectly executed.
There are some major problems with both of
these approaches:
Quick ship — Although this approach is fast and
incremental, two things that we love, it often
leads to parts of the original product design
and vision being left on the cutting room floor.
It’s also hard to know when something is technically “done.” Since it’s being built in small,
9 • Burndown
tiny parts, it introduces new product states
that may not have been accounted for in the
original design.
Big Bang — As the counterbalance to the quick
ship releases, big bangs are usually slow, long
processes that rarely end up as originally intended. Feature and scope creep takes over
and the customer is left out of the process for
long periods of time. No visible progress is being made in the eyes of the customer, which
leaves the business with a lack of feedback on
the decisions they’re making. In addition, the
time estimates of these releases are almost
always underestimated and can drag on for
months later than originally planned and budgeted.
With both of these approaches, we end up
spending a lot of time playing telephone with
each other internally. On the big bang side of
the spectrum, things are constantly changing
mid-flight, which leaves customers, engineers,
designers, execs, marketers, PMs, sales (okay
you get the point)…it leaves them all feeling
like the product is under-delivering in one way
or another. On the other hand, with the quick
ship approach, all of the progress and organi10 • Burndown
zation for next steps tends to live in the head
of the PM. Milestones are unclear and progress
is hard for the rest of the organization to see.
A single source of truth is often lost amongst
all the quick changes and long cycles because
information about different releases winds up
scattered amongst all the different tools PMs
use: Jira, Trello, Prod Pad, Slack, internal wikis,
etc. We all know how messy it can get.
At Drift, we’re always looking for ways to craft
better processes, and over the past year we’ve
developed a new approach to building software
and product.
We call it Responsive Development. It breaks
your product development out of the existing spectrum and lets you move more quickly,
while increasing your ability to focus on the
customer.
We believe Responsive Development matches
the way people want to live and work, and that
the outcome is a better, more useful, and more
reliable product that customers adore.
What are the first principles of Responsive
Development?
11 • Burndown
• It’s flexible: It’s based on principles, not rules.
Rules are binding. Principles favor progress
and forward momentum.
• It’s customer-driven: Always gather firsthand feedback, not second-hand feedback.
Get engineers and designers talking directly to
customers.
• It’s iterative: The first attempt is always
wrong. You must iterate towards the best possible solution, not try to get it on the first try.
• It’s rapid: Speed gives you more iterations
and more opportunities to learn.
• It’s visual: Plans, progress, and updates are
done with screenshots and visuals so everyone
can see what the end result will look like.
• It’s transparent: Everyone at the company
has access to see what’s underway.
• It’s incremental: The power of progression
and step-by-step improvements lead to better
results.
12 • Burndown
• It’s always evolving: Learning is never done,
this is a living framework and will change over
time.
• It favors data over internal opinions: Results
should be measurable and more highly regarded than internal opinions.
• It’s focused: Iterations should be focused,
self-contained, and have as few dependencies
as possible.
13 • Burndown
Chapter 2
The Burndown
Framework
Burndown is the framework that Drift uses to
employ the first principles of Responsive Development.
For example, we use Burndown to make sure
we’re always gathering customer feedback on
an ongoing basis. We feel the customer and
end user’s pain and make sure everyone on the
team (including engineers and designers) are
talking to customers on a daily basis to get that
first hand feedback. We measure customer
adoption, usage, & retention on a daily basis.
We believe that an iterative model that evolves
quickly based on feedback wins. The speed in
14 • Burndown
which we’re executing matters and Burndown
encourages keeping that bar high and never
settling for less. When you look at how easy it
has become for competitors in a crowded market to quickly copy one another, you realize
how vital speed truly is.
Because consistent reprioritization is key, we
try to ensure that at any given moment we’re
building only what will have the biggest impact
on our business’s key results over time.
Most importantly, flexibility is more important
than stability when it comes to execution. The
needs of the customer and the market change
rapidly. The framework is flexible to account
for constant changes.
Since Scrum is the prevailing method for software development these days, let’s see how it
compares to Burndown:
15 • Burndown
We’ve been using the Burndown Framework at
Drift on a daily basis for over a year now and
we’ve found the results to be spectacular (and
our customers feel the same, too). As a result
of Burndown, our team is constantly focused
on the customer and we’ve developed a culture
where speed matters and flexibility is a foundation of how our business operates.
Is Burndown Right for My Team?
If you’re like us, you put your customers first.
Always. At Drift, we’re building tools to make
their lives easier, and we want to make sure we
always stay laser-focused on that goal.
16 • Burndown
In order to implement Burndown, you must
have buy-in from everyone who is on your
team. They need to see the value in speed and
flexibility over rigid plans.
It’s important to keep in mind that things like
roadmaps and binding sprint cycles that come
along with a framework like Scrum only solve
for internal process. Burndown, on the other
hand, is specifically built to match what the
customer wants. We’re willing to sacrifice the
month-by-month stability that comes along
with something like Scrum for the speed and
flexibility that comes with Burndown. (And you
should too.)
If you work in a large, corporate environment,
adopting Burndown can be tricky (if not impossible). Gonna be brutally honest here…the
more process your company already has in
place, the harder it’s going to be to switch.
Companies and teams that would be a good fit
for Burndown have the following traits:
• You truly care about your customers and consider yourself a customer-centric business.
17 • Burndown
• Your product teams are small and focused,
no more than 6 members when including engineers, designers, and a PM.
• You’re building a software product.
• You, as a leader within your team (no matter what role), fundamentally connect with the
first principles of Responsive Development.
• Your company and your product teams value
learning and finding new ways to become more
productive.
Throughout the rest of this book, I’m going to
talk about the gritty details of the tools and
processes you need in order to actually implement the Burndown Framework and take advantage of all that Responsive Development
has to offer.
18 • Burndown
Chapter 3
Getting Started with
Burndown
As the Burndown framework is focused around
speed, prioritization, and flexibility, the best
tool we’ve found to manage the process is Trello.
Burndown is a fundamentally visual framework:
Images and screenshots are at the forefront
of feature development and serve as a way to
communicate exactly what will be done in each
product release.
Each product team (which ideally consists of
one Tech Lead, one dedicated designer, one or
two engineers, and one PM) has an individual
Trello board. More on that later.
19 • Burndown
If we were to build a new version of the Drift
customer dashboard – lets call it “Dashboard
2.0” – we’d have two options:
• We could ship the entire thing at once and
take two weeks to do that, or…
• We could ship five different incremental updates to customers in two-day bursts.
Which would you choose? To us, it’s pretty obvious. Number 2 is the winner. We call those
short bursts “microsprints.”
Each list in the Trello board is a microsprint.
Each microsprint is based on semantic versioning by area of the product. Microsprints
are an iterative piece of a larger feature that
can be shipped independently. They are contained pieces of work that could take anywhere
from 30 minutes to 2–4 days. If anything is going to take longer than that, it needs to be revisited to see if it’s worth tying resources up in
something for such a long amount of time (yes,
4+ days for a single new feature or product update is a long time in the world of Burndown).
20 • Burndown
The Trello board is organized from left to right.
Next-up releases are the furthest left on the
Trello board. As you scroll to the right, you begin to see future releases that are still being
explored and designed.
In this example below, the upcoming microsprints were focused around onboarding and
settings.
NOTE: For a while when we were developing this framework, we relied on versioning of each release (ie: release
x.3.x), as shown above. Over time, we realized that the upkeep of this was causing the team to stress too much about
maintaining number consistency, so we moved away from
that as part of the Burndown process. If you’re concerned
heavily with the ability to historically have a record of
what was done when, you may want to consider maintaining versioning.
21 • Burndown
At the top of each list, there is a HIGH LEVEL
card that explains exactly what is going to happen with that release, written in the format of
a Jobs To Be Done statement. The HIGH LEVEL
card also contains a link to the Sketch file that
has all of the designs for that release.
For those unfamiliar with Jobs To Be Done
(JTBD), the goal of JTBD is to focus on the intended outcome, rather than the actual product or service itself. An example is that a customer may buy a power drill, but the real thing
they want is a hole in their wall to hang a painting.
The Job To Be Done phrasing of this situation,
from the perspective of the customer, would
be “When I want to decorate my apartment, I
need a way to drill a hole in the wall so that I
can hang a painting.
Using this system lets anybody in the company,
at any time, see exactly what the outcome of
each microsprint will be. Transparency is key
to the Burndown framework because priorities are constantly shifting and everyone (execs, sales, success, support) needs to have the
22 • Burndown
ability to keep up with what’s coming down the
pipeline next.
Each HIGH LEVEL card has an engineer who
owns the card. That engineer is responsible for
playing quarterback on the release and working
with the rest of their team (and other teams) to
make sure all dependencies are accounted for
and the end result is as expected. Engineers
assign themselves to HIGH LEVEL cards when
they’re actively working on features.
As you start scrolling to the right on the Trello
board, you’ll come across red labels, which are
used to signify that design work is still needed
for that list. Designers work from left to right,
focusing on the left-most list with the red labels. Once design is done, there’s a conversation between the designer and whichever engineer is likely to work on the feature to make
sure everyone is on the same page with the
intended end-result. PMs are usually not present during these meetings. The designer then
crafts the HIGH LEVEL card, based on the
JTBD that the PM wrote in an “artifact” (an inprogress collection of research...explained in a
later chapter) and breaks that out into individual cards within the list. They remove the red
23 • Burndown
labels, and move on to the next list to the right
with a red label.
Managing these lists in Trello allows us to
be super flexible with customer feedback. As
first-hand feedback comes in, the team is empowered and encouraged to re-consider the
prioritization of the upcoming lists. It’s an ongoing conversation. Even though I, as the PM,
might put X before Y on Tuesday, it’s totally fair
game for a designer or engineer to make a case
on Wednesday as to why the order needs to
be swapped. Then, it’s as easy as dragging and
dropping a list on your Trello board. Responsibility is shared.
Labels we use
There are only a few labels that we have on our
Trello cards. Here they are…
24 • Burndown
Design and engineering are responsible for
the details
Product management’s role is to gather vast
amounts of customer feedback and great examples of how others solve the challenges
that customers face today. This means that
the product manager is defining the high-level
goal (the JTBD) of each list, rather than trying
to figure out all of the design and engineeringfocused details. They’re responsible for filling
the list with screenshots of customer feedback,
market research, artifacts (explained later), and
more context for the rest of the team to utilize. Those are up to those teams to figure out
together, which increases ownership on the
product teams.
As an example of this, I, as a product manager,
would create a column for the high-level need
25 • Burndown
of the list. Let’s use a specific report as an example.
Let’s call this column “Reports 2.0.0”.
In the description of that card, I’ll put something along the lines of:
“Now that we are rolling out a new messaging
style, we need to make sure our reports can
support the data and results from that message. These reports need to handle displaying
percentages, raw numbers, a chart, and profiles.”
In addition to that, in the comments on that
card I will post notes from conversations with
customers, screenshots of feedback they sent
via in-app chat, and some examples of other
products that do an amazing job with this sort
26 • Burndown
of feature. The idea
here is volume, not
curation.
Once the designer
reaches this column
as the next thing
they’re tackling, they
sit down with product
management to talk
through the high-level needs on the card
and any potential solutions based on the
feedback and data
that the PM has collected and added to that card over time (AKA:
“Artifacts” - discussed in a later chapter). From
there, the designer will check in with engineering to get a sense of the limitations and
constraints and circle back with the PM if necessary.
After talking things through with engineering
and product management, the designers take
a first pass at a solution, which they then turn
into an Invision file with a clear walkthrough of
the update. That Invision file is shared inter27 • Burndown
nally, as well as with a handful of random customers as well as ones that have requested it in
the past.
If the reviews are overwhelmingly positive, design will sweat through any details with engineering and break out the HIGH LEVEL card in
the Trello list. Each of these Trello lists is what
we’d consider a microsprint - something shippable on their own that adds value to the customer. If it doesn’t add value to the customer, it
can’t be considered a standalone microsprint.
If, on the other hand, there was a lot of constructive feedback, they will take another pass
and repeat the process. For a major release,
such as rebuilding reports entirely, this cycle
may take 2–5 days. The important thing to note
here is that this does not mean that engineering will implement the whole design at once.
That’s the point of microsprints: They’re iterative.
Breaking out the HIGH LEVEL card
Once design and engineering are happy with
the designs and we have a thumbs up from the
customers, the designer will then work with
engineering to break out the HIGH LEVEL card
28 • Burndown
into individual cards. The designer is responsible for making sure each microsprint is a cohesive step towards the full end result that can
have a solid UX on its own, and the engineer
is responsible for making sure each microsprint is attainable and can be shipped on its
own within a max of a few days. As mentioned
earlier, the HIGH LEVEL card will contain the
overall sketch file, but then each piece of the UI
or functionality that is in that release will get
its own card with an attached screenshot. This
is important because it makes sure that small
things don’t fall through the cracks and all the
details are thoroughly considered.
If the feature release is a larger one, the HIGH
LEVEL card will be broken into multiple different lists so that each list is still a microsprint.
This is one of the things where Burndown is
highly different than something like Scrum.
Microsprints are small, incremental releases
that work as stand-alone product updates.
This leaves room for other higher-level priority work to happen in the midst of reaching the
fully expected UI, while also making sure that
things aren’t pushed aside and forgotten about.
As engineers complete different pieces of a mi29 • Burndown
crosprint, and the individual things are shipped
to production before the entire release, the individual cards are archived. Once an entire list
has been shipped to production, the engineer
leading the HIGH LEVEL card then archives the
entire list and moves on to the next thing.
How lists are changed, updated, and prioritized (and who owns what)
Engineering is usually working on the first 2–3
lists depending on the team size. Individual
engineers should be assigned to no more than
one HIGH LEVEL card (AKA one list) at a time,
as assignment is used to signify who is working
on things at that moment.
At any given time, design is about 5–15 lists
ahead of engineering. If they’re consistently
further ahead than that, it means you should
reconsider the size of your releases or how
you’re prioritizing. Just because they’re ahead
doesn’t mean engineering should be imple30 • Burndown
menting everything they’ve done so far, in that
order. That goes against the fundamental principles of Burndown.
Product management is responsible for owning the priorities of lists 5+. They are working
closely with design on 5–15 to make sure the
right things are up next based on customer
(and business) needs. Design mainly owns those
cards while fleshing out the final design and
breaking out each release into smaller pieces.
Throughout this process, they are validating
the designs with customers (unless the update
is extremely minor).
Product management is spending the majority
of their time on lists 15+ to validate the hypothesis of what should be worked on in the coming days/weeks. To the far right of the Trello
board is where cards are consistently added,
archived, and updated. It’s the breeding ground
for what’s coming down the pipe and needs
more customer validation.
The most important part of all of this process is
that it A) removes the need for a project manager and B) removes the need for the product
manager to continually play project manager
31 • Burndown
throughout this process. The product manager is a knowledge worker and should be doing
things like talking to customers more than they
should be doing things like figuring out every
detail of how things will be solved and implemented.
Nothing is set in stone, ever
There are cases when we get customer feedback or make a strategic shift that re-organizes
our priorities, including cards 1–5. While it may
seem chaotic, and it sometimes is, the order
of the lists change every single day to keep us
working on what matters most. This is the most
fundamentally important aspect of Burndown.
The result is a constant focus on the most
high-leverage items possible, which enables
the team to ship maximum value to customers
that align with business goals.
32 • Burndown
Chapter 4
How to Organize
Product Teams
While it’s not required to organize your teams
a certain way in order to use Burndown, we
wanted to share our philosophy on ideal team
structure.
There are a few key principles behind the foundation of our product teams:
• Each team owns a customer-facing portion
of the product. Each team must have something they can point to and say “that is what
the customer sees as a result of our collective
work.” These teams need to be able to ship customer-facing products without any dependen-
33 • Burndown
cies on any other teams.
• While splitting by code base is nice, it’s not
always ideal. We prefer to have teams crosscollaborating on a code base if it means that
each team can ship independent value. Splitting teams by code base can do wonders to
encourage code ownership. This breaks if, for
example, you have only one team working on
backend - that team then becomes a resource
fought over by other teams and can become a
bottleneck for shipping product updates.
• Some roles will be on multiple teams as you
scale (and can help you scale) Your first PM
might wind up on three teams and stretched
thin until you make your next PM hire. That’s
okay. If you spin out a new product team, a designer might start working across two different
teams. That’s also okay. It helps the organization scale and gives you clarity into where to
make the next hire based on current or foreseeable bottlenecks.
The ideal team makeup consists of the following dedicated resources: one product manager, one designer, one tech lead, one front
end engineer, and one backend engineer. This
34 • Burndown
makeup allows teams to be nimble and each
team member to have clear ownership on their
part of the team’s product. Note: At scale, an
ideal team also includes a dedicated product
marketing manager.
The Ideal Product Team
Product Manager Responsibilities:
• Understanding the market
• Framing the problems the team is solving
• Guiding the overall prioritization
Designer Responsibilities:
• Crafting the solution alongside the PM & Tech
Lead
35 • Burndown
• Iterating on the solution with customers
• Making sure the UX of what is delivered
matches what is intended
Tech Lead Responsibilities
• That the solution delivers upon solving the
problem
• That the engineering team is shipping
• That the product works
• Building new features, fixing bugs
Front End Engineer Responsibilities
• Building new features, fixing bugs
• The things they ship work
Back End Engineer Responsibilities
• Building new features, fixing bugs
• The things they ship work
Product Marketing Manager Responsibilities
• Being the bridge to marketing and sales
• Crafting content and collateral to launch and
market the solutions the team builds
36 • Burndown
Chapter 5
The Design Process
Nailing the design process in Burndown is an
evolving goal. We’ve gone through a few iterations in the past to get to where we are today.
It’s taken a lot of grinding to learn the best flow
here. It’s also important to note that this process may not fit perfectly with your own team’s
structure and can depend on both the type of
designer you hire and the mindset of your Tech
Leads.
There are three stages of the design process
within Burndown (took some lessons from
Dropbox’s process here):
37 • Burndown
Stage 0 - Defining the Job and defining success. This is where we (the product managers)
frame the Job To Be Done (JTBD) that we’re trying to solve for and the metrics we associate
with it being successful once launched. It’s also
when the product manager, designer, and engineer sync up to look at the artifacts the product manager has been creating to understand
what the technical limitations might be.
• A good example of a Job To Be Done is: “When
a lead comes to my website, I want them to
talk to the most relevant sales rep (or a CSM/
Support person if they’re already a customer),
so that I can give each lead the attention they
need, in a timely manner, and close them as
customers.”
• A bad example of a Job To Be Done is: “Add
lead routing into the product to allow people
to route new chats by geography or other attributes.”
• The good example is product agnostic. It
doesn’t matter if Drift is creating the solution or not. This is simply what the customer
is trying to achieve...the job they’re trying to
get done. The bad example defines a solution
38 • Burndown
before acknowledging the actual intent of the
customer.
Stage 1 - Visually solving the job. This is when
design takes what they understand of the
problem, uses the customer research, talks to
customers themselves, and starts to create a
visual solution for the job we’re solving. This
process may go through a few cycles of internal and customer iteration. The bigger the update to the product, the more iterations it’ll go
through.
• During these cycles, there will be at least one
Design Crit, where a fellow designer from another team looks at the solution the designer is
crafting and gives feedback.
Stage 2 - support & QA. This is when the solution is actually being implemented. The designer and the engineers building the solution
spend time together to make sure what’s being implemented: adds value to the customer,
solves the JTBD agreed upon, and works as intended from a UX/UI perspective.
• Kick-off: This stage always starts with a Kickoff. This is a 5-10 minute sync up between ev39 • Burndown
eryone who will be implementing the solution
(the engineers, the PM, and the designer) to
make sure everyone knows exactly what they’ll
be doing. We once had a problem where people would start implementation without fully
understanding the exact details of what they
were trying to accomplish. The Kick-off surfaces any friction or unanswered questions.
We go around the room and each person says
“I’ll be doing X, Y, and Z for this.” The goal of
this meeting isn’t to assign work or to discuss
the details - that should be done in Stage 1. The
goal of this meeting is to verbally make sure
everyone is on the same page about what happens next.
It’s also important to note that the designers
always start by designing the most optimal solution. Once they have that and the team has
agreed on it, they will sit down with the engi40 • Burndown
neer who will be implementing the solution to
figure out how to break it into microsprints.
I’d like to give a hat tip to Elyse Bogacz, who is
our rockstar design lead that has helped nail
down this process over hardcore iterations the
last 6-12 months.
41 • Burndown
Chapter 6
Customer Research &
Creating Artifacts
As noted in Stage 0 of the design process,
it’s important for the team to agree upon the
job that is being solved. Since it is the product manager’s job to frame the solution and to
collect customer feedback, product managers
need to be creating artifacts.
An artifact is simply documentation on a the
problem, market research, a collecting of
pointed feedback from customers, or possible suggested solutions. Often, these take the
form of a collection of screenshots, lists, Job
To Be Done statements, wireframes (pen/paper, whiteboard, software...doesn’t really mat-
42 • Burndown
ter), flow charts, etc. It is what helps define the
scope of a potential V1 for solving the job to be
done.
In addition to dumping any/all feedback related to a specific topic in a Trello card to the far
right of the Trello board, the product manager
should be creating larger artifacts. At Drift, all
of those are posted in an internal Wiki (we use
Confluence) and are shared in Slack with the
43 • Burndown
team well before you actually enter Stage 0 of
the design process. You could also create these
in Google docs or other documentation tool
you use. The important thing is that it’s somewhere that’s easily referenceable and accessible in the future when you do get to stage 0.
Artifacts often look different for different undertakings. There is no “right way” to build an
artifact, so long as it’s useful to the team.
Ideally, before anything enters Stage 0 of the
design process, the product manager will have
created and shared 2-3 different artifacts that
the team has seen and commented on.
The goal of the artifact is to create more context for the design and engineering teams when
they sit down to craft the perfect solution. It
may have suggested solutions, but it is never
what is usually known as a spec or requirement
document.
An artifact takes the perspective of “here’s what
we know, here’s the data, here’s the frame of
the problem, here are a few possible solutions,
here’s what competitors are doing...” rather
than your typical perspective of “The solution
44 • Burndown
must include X, Y, and Z. Here is a wireframe
of said solution, etc.” that a normal spec or requirements doc has. We don’t like those -- it
is not the product manager’s job to create the
solution. It is their job to frame the problem
and provide the highest amount of firsthand
feedback and context as possible so the rest of
the product team can own and implement the
solution.
45 • Burndown
Chapter 7
Dealing with Bugs
Since we’re dealing in contained, continuously
shipped microsprints, it’s much easier to avoid
major bugs in the code because you’re being
gut-checked in production on basically a daily
basis. If bugs do arise, they should be relatively easy to tackle because you’re likely to know
what caused them, unless they go unnoticed
over the course of many microsprints and surface way later.
We believe bugs should be entirely owned by
the engineering team. As such, Github issues
are where they live. When bugs pop up, anybody on the team can create a Github issue. If
46 • Burndown
it’s a customer-reported bug, we give it a red
“Bug” label in Github and post a link to the conversation where it was reported. We prioritize
these bugs over others.
As a general rule of thumb, we’re spending
about 20% of our time tackling bugs, and the
other 80% on new features, infrastructure
work, and improvements to existing features.
As bugs surface, it’s up to the engineer’s discretion which need immediate attention and
which we can hold off from for a bit. This is
why the “Bug” label in Github is so important:
because it increases visibility and urgency into
the fact that it’s happening for a paying customer, which could result in churn if the bug
is critical.
47 • Burndown
While we were implementing Burndown, for
the first few months, once a week, the product
manager on each team would sit down with the
tech lead and go through the backlog of bugs
in the Github repos that their team owns. This
way, the PM and the engineers could get aligned
on what was most important, why, and learn to
trust each other’s judgement over time.
As the Burndown principles and framework
became more a piece of the culture, and as
the PM would trust the engineers to make the
same calls they would, that meeting was no
longer necessary on a weekly basis. Now, at the
start of each week (usually over the weekend)
the tech lead and engineers will go through the
Github issues and pick out the top most important bugs they plan to tackle that week.
Since Burndown doesn’t operate in hard weeklong cycles, there are consistently bugs that are
introduced that require us to halt what we’re
doing and tackle them right away. So we make
sure to leave extra capacity to handle those as
they come up so they can be worked on in parallel.
48 • Burndown
Recently we received a few questions about
what our QA process looks like. As mentioned
in the earlier chapter about the design process,
QA is done by the engineer, designer, and PM
before the product is shipped to production.
Often it’s as simple as the engineer saying “Hey
Amanda [the designer], can you come over
here and take a look at this with me to make
sure it looks and works as expected?” In addition to that, we use Github and engineers share
their pull requests for somebody to take a look
at before it’s pushed.
We’ve looked into many of the automated QA
systems out there, and most of them are excessively expensive and cause your entire
shipping process to slow down. Since one of
Burndown’s edges is speed, this is catastrophic
to the flow. In addition, we’ve seen that using
these systems often alleviates the engineers
from truly owning their code. It’s way easier
for them to feel like “ah, whatever, somebody
else will catch it if it’s broken” rather than being fully responsible for making sure it works in
the first place. No, we don’t write unit tests, but
we expect engineers to own the code they’re
building to make sure it works.
49 • Burndown
In addition, we’ve used other automated frontend testing frameworks, and we’ve found that
often they’re far too brittle and the upkeep of
those actual tests is more than effort than expected, especially as the product surface area
extends rapidly.
We do a few additional things around QA.
Here’s our suite of tools to avoid bugs in the
first place:
• We’ve baked QA into the design process
• We use Assertible to test our endpoints
• We’ve begun using Segment’s Nightmare
framework for critical tests (The onboarding
works, login works, etc). The great thing about
this is that we can actually hook these tests
into our build process (using Travis) and fail the
build when the test fails
• We have a QA tester that we hired via Upwork. He is on-call always and does daily live
tests of critical functions, as well as monitors
our #shipyard room in Slack where we post
all of the things the engineer ship, as they ship
them to test anything new.
50 • Burndown
Chapter 8
Tech Debt
Alongside the conversation about tackling bugs
inevitably comes the question of “well, what
about tech debt? When and how do you tackle
that?”
There are a few things to keep in mind here.
The most important is that the customer does
not care about your tech debt. At all. It is irrelevant for them. Tackling tech debt is often a
deep, dark rabbit hole. Something that the engineers pitch as a 2-day thing inevitably turns
into a 3-week re-write of the entire infrastructure. We’ve all been there.
51 • Burndown
Here’s our rule about tech debt - It is only tackled when tied to customer value. Tech debt will
not be touched until it is tied directly to a new
feature that is being rebuilt. An example of
this might be that the onboarding flows were
hacked together 6 months ago and the code
base is super janky. Let’s say there are small
tweaks you want to make here and there. Of
course, the engineer is likely to want to rebuild
the whole thing because they’re playing with a
jenga set, but changing one step isn’t actually
adding much customer value. In this case, we
wouldn’t tackle any onboarding-related tech
debt unless we were to completely revamp the
onboarding flow to be more useful.
52 • Burndown
Chapter 9
Managing your product
backlog
By now you might be wondering: What do you
do about your backlog that has ideas for 4, 6,
and 12 months out? Where do you keep all of
that information and keep it organized?
Depending on the size of your backlog, the
Trello board you use to implement the Burndown framework may begin to break from too
many lists on the far right of the board, which
is completely fine. That’s what happened to us.
So we evolved.
Now, we use the main product Trello board
only as a place where we scope and document
53 • Burndown
microsprints that will actually get done. If
something’s been in there for over a month and
keeps getting pushed around and never done,
that means it’s not top priority and likely will
never be. So we archive that list.
The Burndown framework is designed to keep
you focused on what’s in the short and near
term around the highest-leverage stuff you
can be doing. Anything designed and scoped
that’s 3+ months out, unless you’re a larger, extremely structured organization, is often going to change drastically by the time you reach
that time anyway. We’ve had a designer on one
of our teams get 3+ months ahead of the team
and ultimately most of what she designed never
made it to the light of day because we learned
quickly that we wanted to take things in another direction. And that’s okay. That’s what we
sign up for using Burndown. The worst-case
scenario there would have been to continue
implementing those designs simply because
they were already done. Since Burndown is
inherently flexible, it allowed us to easily say
no and take a different course that had more
promise.
If you’re like me and are customer-develop54 • Burndown
ment focused, you still need a place to organize all of the customer feedback you get on a
weekly and daily basis. I’d suggest either making a different Trello board to hold all of these
ideas and pieces of feedback or keep expanding the width of your Trello board indefinitely
so long as you go through and gut-check it every few weeks and archive things that are not
aligned with where the product is going.
55 • Burndown
Chapter 10
Prioritization
The best way to determine what the next most
important list to tackle is to ask yourself this…
“What is the one thing that provides the most
value to our customers and aligns with our business goals and focuses?”
Answering this question should provide you
with the thing that will give you the most leverage in the long-term to continue adding value
to your customers. The question is a derivative
of the question that Gary Keller poses in his
book, The One Thing, which is “What’s the one
thing I can do right now to make everything
56 • Burndown
else easier or unnecessary?”
We’re big fans of this philosophy at Drift as we
believe it helps us focus on the right things.
We also thematically structure the upcoming
priorities of the company, which we put into
a single slide or list and repeat over and over
again to the team. In our case, an example list
would look something like this…
• Keeping the product stable
• Driving more signups
• Increasing the activation rate
• Innovating our chat widget
This gives us all a guideline to reference when
we’re asking ourselves what we could be focusing on next. If you can’t do something that
directly benefits #1, then work on something
that benefits #2, etc.
The power of microsprints really shines when it
comes to prioritization. Instead of locking ourselves into, let’s say, two full weeks of work on
a new dashboard, we can commit to five twoday microsprints. After we get through the first
two microsprints (that add customer-facing
value on their own), we may shift priorities as
57 • Burndown
a business around one core goal. Let’s say we
learn that what we’ve been building isn’t useful or we simply come to recognize that working on something like billing gives us way more
leverage. Because we broke things into microsprints, we can easily table the next three microsprints for the dashboard and instead focus
on that new high-leverage work, thus coming
back to the dashboard later.
58 • Burndown
Chapter 11
Transparency &
Communication
Since the Burndown framework is one that
constantly shifts as the microsprints inside of
it are continually moving to match the needs,
wants, and desires of the customer, it’s important to make sure the team is on track and has
milestones to hit.
Burndown doesn’t subscribe to the concept of
a roadmap with deadlines. We fundamentally believe that roadmaps only serve company
needs, not customer needs.
So here’s how we stay aligned on priorities…
59 • Burndown
Every Friday, the product manager will sit
down with the tech lead of each team (or the
entire team, depending on that product team’s
culture) and talk through the upcoming microsprints for about 10–15 minutes. Sometimes,
it’s clear what the next most important thing
is and we’re immediately on the same page.
Other times, good points will be brought up
in the conversation and microsprints/lists will
be shifted a bit and agreed upon as the focuses
for the upcoming week. During that meeting,
the product manager writes down exactly what
each engineer and the designer is committing
to for that upcoming week. Then they’ll all review that list before leaving the room to get in
agreement.
Each Friday afternoon, the product manager
on each team then writes an internal wiki post
with a tactical breakdown of which lists from
Trello will be focused on during that upcoming
week and what value they provide to the customer. That gets posted on Sunday evening in
Slack. This internal post generally takes about
an hour total to write, but it’s 100% worth it to
keep the team aligned as they show up the next
morning. If done right, this is the only project
management that a product manager would
60 • Burndown
need to do on a weekly basis.
Here’s the structure of those posts (one per
product team each week)…
1. A brief recap on the week prior, a list of
what the team shipped the previous week,
and a high-level list of the themes that are
being focused on this next week
2. A breakdown of what each individual on the
team will be doing that upcoming week
3. Visuals/screenshots (from the HIGH LEVEL
cards in Trello) of what will be shipped
4. A visual representation of the upcoming
themes
At the top of this internal post (#1 in the list
above), there will be a few sentences on what
themes were completed last week, and a rough
outline of what’s coming next week. Here’s an
example:
61 • Burndown
Then, for part #2, the product manager takes
the things that were agreed upon in the Friday
meeting and puts them into a table. You’ll notice that each person on the team has a main
focus -- that’s their one thing that’s more important than anything else on their list for that
week.
62 • Burndown
For part #3, each of the main focuses is broken
down alongside the Job To Be done and a link
to the microsprint in Trello. Since Burndown is
visual, screenshots of what will be released in
that microsprint are added for everyone to see
what (in an ideal world) will be in the product
sometime over the course of the next week.
Part #4 is a high-level overview of the entire
product organization’s upcoming priorities.
This is for the execs, sales, success, and marketing teams to see what’s roughly coming
down the pipeline at a higher-level. The below
example theme document is from a customdesigned Sketch file. Once you scale past a
few product managers, you’ll need to look into
managing this off of a collaborative tool such
as Google Sheets or Roadmunk. That way it can
be linked to at all times and you don’t have to
worry about multiple PMs trying to edit/up63 • Burndown
date this document at the same time.
Remember, within each week there can be
many microsprints planned and we are shipping things immediately when they are completed. These weekly check-ins and updates
serve more as a way to ensure that everyone
has a sense of what progress looks like and a
tangible means to know exactly what to focus
on when they get in on Monday.
The things posted in the Sunday update aren’t a
64 • Burndown
“this must be done no matter what” kind of list.
Sometimes things will happen and things we
planned to do are no longer worked on since
something higher priority comes up that could
better impact our current company priorities/
themes. That’s the beauty of the flexibility of
short microsprints.
65 • Burndown
Chapter 12
Getting Your Team Onboard
If you’re a larger organization with a strongly
rooted process, I’d recommend working within
one product team to start (a couple engineers,
a PM, and a designer).
Get one team working well with microsprints
and semantic versioning in Trello, prove it’s an
efficient way to build products, and then scale
it across your organization.
If you’re just starting up or building, you should
start with microsprints in Trello and bake it
into your product culture. Semantic versioning
is like a perfect set of training wheels as you
66 • Burndown
and your team get started with microsprints,
so take them seriously and stick with them.
When you feel like they’re truly getting in the
way of your speed, that’s when you know they
are no longer necessary. For reference, it took
us about 3-4 months to get there.
Here’s how our semantic versioning system
looks...
• N.x.x (Major Release) - A major release is
when a brand new part of the product is introduced or an existing part of the product gets a
complete overhaul.
• x.N.x (Minor Release) - A minor release is
when a new feature is introduced within an existing part of the product.
• x.x.N (Patch) - A patch is when an existing
part of a feature is tweaked slightly.
In addition, here’s a helpful set of common objections you might get from the rest of your
product team if you’re trying to move to Burndown alongside suggested responses…
The objection: We already have a process that
67 • Burndown
works. Why would we change it?
Potential response: If you told me that you
had $10 to spend and you were going to store A
and instead I told you that right next door we
could go to store B where your $10 can buy you
twice as much, would you follow me to store
B? Probably. Ultimately, the incremental and
weekly benefit we get from switching to Burndown will pay for itself in only a few weeks and
we’ll be delivering more value to our customers
more often.
The objection: How do you know this is going
to work?
Your response: It’s been tested and refined by
David Cancel, five-time founder, and the product teams at Drift, HubSpot, and many others.
They’ve surfaced challenges and have a clear
framework for how to implement it on our
team. There’s also an entire handbook on the
process that you can check out first.
The objection: How long will it take us to get
this new process up and running?
Your response: About the length of one full
68 • Burndown
two-week sprint. We’ll move all of our backlog
into Trello boards over the next few days and
we can even start using it today for the things
that the product teams are currently tackling.
The objection: There are details I think are
missing from this framework and questions I
have. How would we reconcile those?
Your response: Like we always do...as a team.
Burndown is inherently flexible and ever-evolving. We can absolutely adapt it to our team and
how we work best. This isn’t a means of saying
“we must operate exactly like this” but more of
a means of saying “this way of building product
seems better. Let’s take the pieces that work
for us and improve our own process with it.”
The objection: I don’t use tools like Trello,
Sketch, Invision, and Github, etc. Are they required?
Your response: You need some sort of tool to
manage microsprints. You could do a real-life
wall with sticky-notes, but it’s not recommended. Almost all of these tools are super simple to
get started with. Trello is also 100% free and
barely has any learning curve at all.
69 • Burndown
The objection: I work in a larger, traditional,
organization where I need to set go to market
milestones throughout the year with marketing, sales, etc. How can I do that while also using Burndown?
Your response: We almost always publicly
launch or release products that we’ve already
rolled out to customers. This allows us to get
case studies and quotes from customers already using the tool, and provides marketing
and sales with a product that they know will
work and that they have validated use cases for.
70 • Burndown
Conclusion
In the world of B2B software, consumer software, mobile apps, SaaS, etc. the reality is the
days when innovative technology was a true
differentiator are coming to a close. Today,
most software companies build products using the same tools as everyone else.
One of the best ways you can set yourself
apart is through the process you use to build
that software. Responsive Development and
the Burndown framework provide a more
customer-focused, flexible model of building
software that can give you and your company
an edge against competitors.
71 • Burndown
At Drift, we’re obsessed with helping every
company in the world know, grow, and amaze
their customers. It only feels right that we
make sure the process we use to build our
tools follows that philosophy.
Thank you for reading this book on Responsive Development please feel free to reach out
to [email protected] if you have any questions
or feedback on this book or Burndown. We’re
here as a resource for you as you implement
Burndown and would love to hear how it goes.
If you enjoyed this book and think some other
people may find it useful, we’d be so grateful if
you tweet out a link to it.
Good luck, from your friends at Drift.
- Matt Bilotti & David Cancel
Matt Bilotti
72 • Burndown
David Cancel
Download